[libvirt] [RFC PATCH v1 1/4] numa: describe siblings distances within cells

Daniel P. Berrange berrange at redhat.com
Tue Jun 27 13:55:19 UTC 2017


On Tue, Jun 27, 2017 at 03:46:09PM +0200, Wim ten Have wrote:
> On Wed, 21 Jun 2017 21:08:30 +0200
> Wim ten Have <wim.ten.have at oracle.com> wrote:
> 
> > On Mon, 19 Jun 2017 12:26:20 -0600
> > Jim Fehlig <jfehlig at suse.com> wrote:
> > 
> > > On 06/12/2017 12:54 PM, Wim Ten Have wrote:  
> > > > From: Wim ten Have <wim.ten.have at oracle.com>
> > > >
> > > > Add libvirtd NUMA cell domain administration functionality to
> > > > describe underlying cell id sibling distances in full fashion
> > > > when configuring HVM guests.    
> > > 
> > > Thanks for the detailed cover letter :-). Some of that information (e.g. ACPI 
> > > spec, SLIT table, qemu support) would be useful in this commit message.  
> > 
> >   I'll follow up on that under v2.
> > 
> > > > [below is an example of a 4 node setup]
> > > >
> > > >   <cpu>
> > > >     <numa>
> > > >       <cell id='0' cpus='0' memory='2097152' unit='KiB'>
> > > >         <distances>
> > > >           <sibling id='0' value='10'/>
> > > >           <sibling id='1' value='21'/>
> > > >           <sibling id='2' value='31'/>
> > > >           <sibling id='3' value='41'/>
> > > >         </distances>
> > > >       </cell>    
> > > 
> > > Others on the list are better XML designers, but the following seems to achieve 
> > > the same and is a bit more compact
> > > 
> > >          <cell id='0' cpus='0' memory='2097152' unit='KiB'>
> > >            <sibling id='0' distance='10'/>
> > >            <sibling id='1' distance='21'/>
> > >            <sibling id='2' distance='31'/>
> > >            <sibling id='3' distance='41'/>
> > >          </cell>
> > > 
> > > CC danpb, who often has good ideas on schema design.  
> > 
> >   Will give Daniel and (others) time prior submitting v2.  Otherwise
> >   follow your suggestion above for its compact and clear annotation.
> 
>   One thing on specific, forgot to mention previous reply, is that the
>   selected syntax was derived from way libvirt describes host capabilities
>   introduced under;
> 
>     commit 8ba0a58f8d9db9eeed003b889dfcd3451d10fbca
> 
>       Author: Michal Privoznik <mprivozn at redhat.com>
>       Date:   Tue Jun 3 15:18:27 2014 +0200
>       "virCaps: Expose distance between host NUMA nodes"

Yep, so I agree with Jim that leaving out the extra <distances>
element nesting would be more concise.  Given, that we already
have the more verbose approach used in host capabilities though,
I think it is preferrable to taken the more consistent approach
you've done.  It is confusing for users when the same thing is
represented in multiple different ways, and NUMA setup is already
very confusing, so good to avoid making that worse :-)

I think your current patch proposal is fine from the XML pov.


Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|




More information about the libvir-list mailing list